Linux 5.8.5: amdgpu: [mmhub] page fault (src_id:0 ring:169 vmid:0 pasid:0, for process pid 0 thread pid 0)

Brief summary of the problem:

Just booted 5.8.5 on my workstation with RX5700XT and dmesg is flooded with the following (appears 1-2 times per second):

[  597.475418] amdgpu 0000:07:00.0: amdgpu: [mmhub] page fault (src_id:0 ring:169 vmid:0 pasid:0, for process  pid 0 thread  pid 0)
[  597.487006] amdgpu 0000:07:00.0: amdgpu:   in page starting at address 0x00000000fffb8000 from client 18
[  597.496499] amdgpu 0000:07:00.0: amdgpu: GCVM_L2_PROTECTION_FAULT_STATUS:0x00041F52
[  597.504171] amdgpu 0000:07:00.0: amdgpu:      Faulty UTCL2 client ID: 0xf
[  597.510712] amdgpu 0000:07:00.0: amdgpu:      MORE_FAULTS: 0x0
[  597.516299] amdgpu 0000:07:00.0: amdgpu:      WALKER_ERROR: 0x1
[  597.521969] amdgpu 0000:07:00.0: amdgpu:      PERMISSION_FAULTS: 0x5
[  597.528077] amdgpu 0000:07:00.0: amdgpu:      MAPPING_ERROR: 0x1
[  597.533840] amdgpu 0000:07:00.0: amdgpu:      RW: 0x1

Not seeing this on my previous kernel (5.7.14). The problem goes away when booting 5.8.5 with intel_iommu=off.

Hardware description:

• CPU: 2x Intel Xeon E5-2673 v4
• GPU: Radeon RX 5700 XT
• System Memory: 256GB
• Display(s): 2x Dell U2410
• Type of Diplay Connection: 2x DP

System infomration:

• Distro name and Version: Gentoo Base System release 2.7
• Kernel version: 5.8.5
• Custom kernel: gentoo-sources
• AMD package version: No package

How to reproduce the issue:

• Upgrade from Linux 5.7.14 to 5.8.5
• Make sure Intel IOMMU is enabled
• Reboot

Attached files:

• dmesg.txt
• Xorg.0.log